Calabi–Yau Compactification and Moduli

A compact internal space determines both the amount of supersymmetry visible in four dimensions and the fields that a low-energy observer can excite. For a smooth Calabi–Yau threefold with full $SU(3)$ holonomy, flux-free type II compactification preserves eight real supercharges. Its massless spectrum is controlled by two Hodge numbers, $h^{1,1}$ and $h^{2,1}$, while the interactions of its scalar fields depend on metrics on the spaces of geometric deformations.

We will derive the connection between these statements. The main steps are to identify the allowed metric deformations, turn their internal wavefunctions into four-dimensional kinetic terms, and reduce the ten-dimensional form fields without double-counting dual descriptions. The final comparison is

$$\begin{array}{c|cc} & n_V & n_H\\ \hline \text{IIA on }X & h^{1,1} & h^{2,1}+1\\ \text{IIB on }X & h^{2,1} & h^{1,1}+1 \end{array}$$

Here $n_V$ counts vector multiplets beyond the gravity multiplet, and $n_H$ counts hypermultiplets. The shared Kähler geometry and internal orientation fix the metric, form and string-unit normalizations. The ten-dimensional fields are those of the type II low-energy actions.

Ricci-flat geometry and preserved supersymmetry

Consider the leading product background

$$M^{3,1}\times X,$$

with constant dilaton, vanishing background form fluxes, and no orientifold projection or branes. The internal space $X$ has real dimension six. For a classical geometric reduction, its curvature radii must be large compared with $\ell_s$, the string coupling must be small, and the four-dimensional energies must lie below the relevant Kaluza–Klein masses. A large total volume does not ensure that every cycle remains large.

An almost complex structure is an endomorphism $I$ of the tangent bundle satisfying $I^2=-1$. Integrability means that its holomorphic directions arise from local complex coordinates; equivalently, its Nijenhuis tensor vanishes. A compatible metric obeys $g(Iu,Iv)=g(u,v)$. Its associated two-form is $J(u,v)=g(Iu,v)$.

For an integrable complex structure, the Kähler condition $dJ=0$ is equivalent to $\nabla I=0$. Locally the Hermitian metric comes from a real function $\mathcal K_{\mathrm{loc}}$:

$$g_{i\bar j}=\partial_i\partial_{\bar j}\mathcal K_{\mathrm{loc}}.$$

Changing $\mathcal K_{\mathrm{loc}}$ by a holomorphic function plus its complex conjugate leaves the metric unchanged. This local potential is distinct from the Kähler potentials on moduli space that we will derive below.

The Ricci form of a Kähler metric is

$$\mathcal R_{\mathrm{Ric}} =-i\partial\bar\partial\log\det(g_{i\bar j}), \qquad [\mathcal R_{\mathrm{Ric}}]=2\pi c_1(X).$$

In components, $\mathcal R_{\mathrm{Ric}}=iR_{i\bar j}\,dz^i\wedge d\bar z^{\bar j}$ with $R_{i\bar j}=-\partial_i\partial_{\bar j}\log\det g$; the order of the two Dolbeault derivatives fixes the sign.

A Ricci-flat Kähler metric therefore requires vanishing first Chern class in real cohomology. Yau’s theorem supplies the converse existence statement: on a compact Kähler manifold with this condition, each Kähler class contains a unique Ricci-flat Kähler metric for the fixed complex structure. The theorem establishes existence and uniqueness; it usually does not supply an explicit metric. Yau’s announcement states this result and its application to projective hypersurfaces.

For the compactifications studied here we also require a nowhere-vanishing holomorphic three-form $\Omega$, and the holonomy of the Ricci-flat metric is exactly $SU(3)$. These qualifications matter. Vanishing real $c_1$ alone does not establish holomorphic triviality of the canonical bundle, and holonomy contained in $SU(3)$ need not equal $SU(3)$. The flat six-torus and $K3\times T^2$ have smaller holonomy and preserve additional supersymmetry.

Why does full $SU(3)$ holonomy give eight supercharges? A supersymmetry parameter contains an internal spinor $\eta$ obeying

$$\nabla_m\eta=0.$$

Parallel transport must therefore leave $\eta$ invariant. The relevant spin group is $\operatorname{Spin}(6)\simeq SU(4)$, the double cover of $SO(6)$. Its chiral spinor restricts as

$$\mathbf4\big|_{SU(3)}=\mathbf3\oplus\mathbf1.$$

There is one complex singlet of each conjugate internal chirality. The ten-dimensional Majorana condition relates the conjugate terms in the decomposition of a supersymmetry parameter; it does not double the number of independent four-dimensional supercharges. Each ten-dimensional type II parameter supplies four real supercharges, and the two parameters supply eight: four-dimensional $\mathcal N=2$.

The integrability condition

$$[\nabla_m,\nabla_n]\eta =\frac14R_{mnab}\gamma^{ab}\eta=0$$

also explains Ricci flatness. Contracting with a gamma matrix and using the curvature Bianchi identity gives $R_{mn}\gamma^n\eta=0$. In a positive-definite internal metric, Clifford multiplication by a nonzero real vector is invertible. Since $\eta$ is nonzero, every vector $R_{mn}$ must vanish. This implication does not say that an arbitrary Ricci-flat manifold has a parallel spinor.

Hodge data and geometric zero modes

On a compact Kähler manifold, differential forms split into types $(p,q)$, and Hodge theory identifies cohomology classes with harmonic representatives:

$$H^k(X,\mathbb C) =\bigoplus_{p+q=k}H^{p,q}(X), \qquad h^{p,q}=\dim H^{p,q}(X).$$

Complex conjugation and duality give

$$h^{p,q}=h^{q,p}, \qquad h^{p,q}=h^{3-p,3-q}.$$

Full $SU(3)$ holonomy supplies further information. On a compact Ricci-flat Kähler manifold, holomorphic forms are parallel. The $SU(3)$ representation on one-forms and two-forms has no holomorphic singlet, while its determinant representation has one. Thus

$$h^{1,0}=h^{2,0}=0, \qquad h^{3,0}=1.$$

Together with connectedness, these facts determine the Hodge diamond from $h^{1,1}$ and $h^{2,1}$.

For a smooth compact threefold with full $SU(3)$ holonomy, only $h^{1,1}$ and $h^{2,1}$ are independent. Each horizontal row has fixed degree $k=p+q$ and sums to $b_k$. This is a cohomology diagram, not a picture of the internal manifold.

Summing each fixed-degree row yields

$$\begin{aligned} b_0=b_6&=1,& b_1=b_5&=0,\\ b_2=b_4&=h^{1,1},& b_3&=2h^{2,1}+2. \end{aligned}$$

Consequently,

$$\chi(X)=\sum_{k=0}^6(-1)^k b_k =2\bigl(h^{1,1}-h^{2,1}\bigr).$$

The physical use of cohomology is easiest to see in a mode expansion. For a minimally coupled scalar, let $\Delta_XY_n=\lambda_nY_n$, with the internal Laplacian chosen nonnegative. Expanding $\Phi(x,y)=\sum_n\phi_n(x)Y_n(y)$ gives four-dimensional masses $m_n^2=\lambda_n$ in the unwarped product metric. Form fields have the analogous expansion in eigenforms, after their gauge redundancies are removed. Their zero modes are harmonic forms.

Metric fluctuations require more care: they obey the linearized Ricci-flat equations modulo diffeomorphisms, rather than a scalar Laplace equation. On a Calabi–Yau threefold these fluctuations divide into $h^{1,1}$ real Kähler deformations and $h^{2,1}$ complex complex-structure deformations. We now construct both kinds and their kinetic metrics.

Kähler deformations and their kinetic metric

Use the dimensionless forms of the shared definitions:

$$j=v^a\omega_a, \qquad b=b^a\omega_a, \qquad T^a=b^a+i v^a, \qquad a=1,\ldots,h^{1,1}.$$

The $\omega_a$ represent an integral basis of the free part of $H^2(X,\mathbb Z)$. The coefficients $v^a$ must lie in the Kähler cone, where $j$ is represented by a positive Kähler form. An arbitrary choice of integral basis does not make this cone simply the region $v^a>0$.

Define the intersection data

$$\begin{aligned} \kappa_{abc}&=\int_X\omega_a\wedge\omega_b\wedge\omega_c,\\ \kappa_a&=\kappa_{abc}v^bv^c, \qquad \kappa_{ab}=\kappa_{abc}v^c. \end{aligned}$$

The dimensionless volume and its first two derivatives are

$$\widehat{\mathcal V} =\frac16\kappa_{abc}v^av^bv^c, \qquad \partial_a\widehat{\mathcal V}=\frac12\kappa_a, \qquad \partial_a\partial_b\widehat{\mathcal V}=\kappa_{ab}.$$

These are topological intersection numbers combined with continuous metric parameters. The physical string-frame volume is $\ell_s^6\widehat{\mathcal V}$.

To obtain a kinetic metric, we need the inner products of the harmonic forms, using the shared dimensionless-metric operator $\widehat{*}_6$. Decompose a harmonic $(1,1)$ form into a primitive part and a component along $j$:

$$\omega_a=\omega_a^{\mathrm P} +\frac{\kappa_a}{6\widehat{\mathcal V}}j, \qquad j^2\wedge\omega_a^{\mathrm P}=0.$$

The coefficient is constant: the contraction of a harmonic $(1,1)$ form with the Kähler form is a harmonic function, hence constant on compact connected $X$. In complex dimension three,

$$\widehat{*}_6\omega_a^{\mathrm P}=-j\wedge\omega_a^{\mathrm P}, \qquad \widehat{*}_6j=\frac12j^2.$$

Substitution gives

$$\widehat{*}_6\omega_a =-j\wedge\omega_a +\frac{\kappa_a}{4\widehat{\mathcal V}}j^2.$$

It follows that

$$\int_X\omega_a\wedge \widehat{*}_6\omega_b =-\kappa_{ab} +\frac{\kappa_a\kappa_b}{4\widehat{\mathcal V}}.$$

The left-hand side is a positive inner product on nonzero real harmonic forms. This is the geometric reason the apparently indefinite intersection expression produces a positive kinetic metric inside the Kähler cone.

The leading IIA reduction, expressed in four-dimensional Einstein frame and at fixed four-dimensional dilaton, gives

$$G^{\mathrm K}_{a\bar b} =\frac1{4\widehat{\mathcal V}} \int_X\omega_a\wedge \widehat{*}_6\omega_b =-\frac{\kappa_{ab}}{4\widehat{\mathcal V}} +\frac{\kappa_a\kappa_b}{16\widehat{\mathcal V}^2}.$$

Here the scalar kinetic term is $-M_4^2G^{\mathrm K}_{a\bar b}\partial T^a\partial\bar T^{\bar b}$, with $M_4$ the reduced four-dimensional Planck mass. Equivalently, the real field-space line element is $2G^{\mathrm K}_{a\bar b}dT^a d\bar T^{\bar b}$. The form-norm coefficient follows from the Einstein-frame reduction; Grimm and Louis, section 2 give the corresponding full leading action.

There is a useful independent check. Since the volume depends only on $v^a=\operatorname{Im}T^a$,

$$K_{\mathrm K}=-\log\bigl(8\widehat{\mathcal V}\bigr), \qquad \partial_{T^a}\partial_{\bar T^{\bar b}}K_{\mathrm K} =\frac14\partial_{v^a}\partial_{v^b}K_{\mathrm K} =G^{\mathrm K}_{a\bar b}.$$

The factor $1/4$ is essential. For one modulus with $\widehat{\mathcal V}=\kappa v^3/6$, the result is $G_{T\bar T}=3/(4v^2)$, regardless of the positive constant $\kappa$.

At large radius this metric also has a special Kähler description through a homogeneous cubic prepotential. In a compatible projective normalization its leading part is

$$\mathcal F(X) =-\frac16\frac{\kappa_{abc}X^aX^bX^c}{X^0}, \qquad T^a=\frac{X^a}{X^0}.$$

The cubic is a classical large-radius result. Perturbative and worldsheet-instanton contributions can modify the exact prepotential. The geometric norm calculation above does not compute those contributions, and it is not a formula for the complete IIB hypermultiplet metric.

Complex-structure deformations and the period metric

Changing the complex structure changes which one-forms are holomorphic. Infinitesimally, describe the new $(1,0)$ coframe by

$$dz^i+\mu^i{}_{\bar j}\,d\bar z^{\bar j}.$$

The Beltrami differential $\mu$ is a $(0,1)$ form valued in the holomorphic tangent bundle. Linearized integrability gives $\bar\partial\mu=0$, while an infinitesimal coordinate change shifts $\mu$ by a $\bar\partial$-exact term. Thus inequivalent infinitesimal deformations belong to $H^1(X,T_X)$.

From the Beltrami differential to the metric

The holomorphic volume form converts the vector index of $\mu$ into two form indices. With the usual component expansion of a three-form,

$$\begin{aligned} \Omega&=\frac1{3!}\Omega_{ijk}\,dz^i\wedge dz^j\wedge dz^k,\\ \chi=\mu\lrcorner\Omega &=\frac12\mu^i{}_{\bar j}\Omega_{ikl}\, d\bar z^{\bar j}\wedge dz^k\wedge dz^l. \end{aligned}$$

Because $\Omega$ never vanishes, contraction is an isomorphism of the relevant bundles and hence of their Dolbeault cohomologies:

$$H^1(X,T_X)\simeq H^{2,1}(X).$$

This explains the number $h^{2,1}$. The extension of infinitesimal Calabi–Yau deformations to a smooth local family uses the mathematical unobstructedness result; the cohomology isomorphism alone is only a first-order calculation.

The relation to a metric fluctuation is also explicit. Expanding the Hermitian metric in the deformed coframe gives its $(0,2)$ component,

$$\delta g_{\bar i\bar j} =g_{k\bar i}\mu^k{}_{\bar j} +g_{k\bar j}\mu^k{}_{\bar i}.$$

For a harmonic Calabi–Yau deformation, $\chi$ is primitive. Indeed, Kähler identities make $\Lambda\chi$ a harmonic $(1,0)$ form, and $h^{1,0}=0$. Primitivity is equivalent to symmetry of the lowered Beltrami tensor in this description, so

$$\delta g_{\bar i\bar j}=2g_{k\bar i}\mu^k{}_{\bar j}.$$

Its conjugate gives $\delta g_{ij}$. The factor two follows from the two metric slots. It would change if one defined the deformation variable as a component of $\delta I$ instead of the coframe coefficient used here.

The inverse relation and its normalization follow from

$$\begin{aligned} \|\Omega\|^2&=\frac1{3!}\Omega_{ijk}\bar\Omega^{ijk},\\ \Omega_{mkl}\bar\Omega^{ikl} &=2\|\Omega\|^2\delta_m{}^i,\\ \mu^i{}_{\bar j} &=\frac{\bar\Omega^{ikl}\chi_{kl\bar j}} {2\|\Omega\|^2}. \end{aligned}$$

In particular, $\|\chi\|^2=\|\Omega\|^2\|\mu\|^2$. A parallel $\Omega$ has constant pointwise norm on each Ricci-flat fiber. The metric-deformation norm therefore reduces to

$$\begin{aligned} \frac1{2\mathcal V}\int_X\!\mathrm{vol}_X\, g^{i\bar j}g^{k\bar l} \delta g_{ik}\delta g_{\bar j\bar l} &=\frac2{\mathcal V}\int_X\!\mathrm{vol}_X\,\|\mu\|^2\\ &=2\frac{\int_X\chi\wedge *_6\bar\chi} {\int_X\Omega\wedge *_6\bar\Omega}. \end{aligned}$$

Here $\mathcal V=\int_X\mathrm{vol}_X$; common length factors cancel in this ratio. For moduli coordinates $u^i$ and a tangent displacement $\delta u^i$, set $\chi=\chi_i\delta u^i$. The right-hand side is then the real line element $2G^{\mathrm{cs}}_{i\bar j}\delta u^i\delta\bar u^{\bar j}$. The label on $\chi_i$ now enumerates moduli; each $\chi_i$ is itself an internal three-form.

Projecting out the normalization of Ω

Choose a local holomorphic section $\Omega(u)$ over complex-structure moduli space, using a fixed cohomology basis to compare nearby fibers. Writing $\partial_i=\partial/\partial u^i$, its derivative decomposes as

$$\partial_i\Omega=k_i\Omega+\chi_i, \qquad \chi_i\in H^{2,1}(X).$$

The first term changes the choice of representative in the holomorphic three-form line; the second measures a change in complex structure. Hodge type makes $\int\chi_i\wedge\bar\Omega=0$.

The decomposition $\partial_i\Omega=k_i\Omega+\chi_i$ separates a change of normalization from the primitive $(2,1)$ deformation. The normalized Hodge inner product of the latter gives the complex-structure metric. The drawing represents directions in cohomology space; angles and lengths are schematic.

Let

$$S(u,\bar u)=i\int_X\Omega\wedge\bar\Omega>0.$$

Holomorphic dependence and Hodge type give

$$\begin{aligned} \partial_i S&=k_iS,\\ \partial_i\partial_{\bar j}S &=k_i\bar k_{\bar j}S +i\int_X\chi_i\wedge\bar\chi_{\bar j}. \end{aligned}$$

The terms along $\Omega$ cancel when we differentiate the logarithm:

$$\begin{aligned} K_{\mathrm{cs}}&=-\log S,\\ G^{\mathrm{cs}}_{i\bar j} =\partial_i\partial_{\bar j}K_{\mathrm{cs}} &=-\frac{\int_X\chi_i\wedge\bar\chi_{\bar j}} {\int_X\Omega\wedge\bar\Omega}. \end{aligned}$$

The minus sign is compatible with positivity. In the shared internal orientation,

$$*_6\Omega=-i\Omega, \qquad *_6\chi_i=i\chi_i,$$

so the same result is

$$G^{\mathrm{cs}}_{i\bar j} =\frac{\int_X\chi_i\wedge *_6\bar\chi_{\bar j}} {\int_X\Omega\wedge *_6\bar\Omega}.$$

This is the Weil–Petersson metric on complex-structure moduli. It is also the leading IIB vector-multiplet metric and appears in the IIA hypermultiplet geometry. The distinction between the geometric metric and the complete hypermultiplet metric is described in Grimm and Louis’s IIA reduction.

Under a nonzero holomorphic rescaling $\Omega\mapsto f(u)\Omega$,

$$K_{\mathrm{cs}}\mapsto K_{\mathrm{cs}}-\log|f|^2, \qquad G^{\mathrm{cs}}_{i\bar j}\mapsto G^{\mathrm{cs}}_{i\bar j}.$$

One must retain this holomorphic-section description during differentiation. Normalizing $S$ to a constant at every point generally requires a nonholomorphic rescaling; differentiating that constant would not calculate the metric.

Periods as local coordinates

Choose a real symplectic basis of $H^3(X)$ with

$$\int_X\alpha_K\wedge\beta^L=\delta_K{}^L, \qquad \int_X\alpha_K\wedge\alpha_L =\int_X\beta^K\wedge\beta^L=0,$$

where $K,L=0,\ldots,h^{2,1}$. Write

$$\Omega=X^K\alpha_K-\mathcal F_K\beta^K.$$

Antisymmetry of the wedge product of three-forms then gives

$$S=i\bigl(\bar X^K\mathcal F_K-X^K\bar{\mathcal F}_K\bigr).$$

The coefficients are periods in a compatible cycle basis. On a patch where $X^0\neq0$, projective ratios $X^i/X^0$ provide special coordinates. The local holomorphic prepotential obeys $\mathcal F_K=\partial\mathcal F/\partial X^K$. Neither a single period gauge nor a single coordinate patch need cover the entire moduli space.

Reducing the type II fields

A four-dimensional $\mathcal N=2$ vector multiplet contains one vector and one complex scalar; a hypermultiplet contains four real scalars. The gravity multiplet includes one vector, the graviphoton. We count propagating bosonic fields around the flux-free background, using the inherited bulk RR potentials. Their coefficient normalizations do not affect the multiplicities below.

The common NS fields give $h^{1,1}+2h^{2,1}$ real metric moduli, $h^{1,1}$ internal $B$-field scalars, one external two-form $B_2^{(4)}$, and the dilaton. A massless two-form in four dimensions is locally dual to a scalar, so the NS sector has

$$n_{\mathrm{NS,scalar}}=2h^{1,1}+2h^{2,1}+2.$$

The four-dimensional dilaton involves the volume: dimensional reduction of $e^{-2\Phi}\sqrt{-G^{\mathrm S}}\mathcal R$ gives a factor $e^{-2\Phi}\widehat{\mathcal V}$. Thus one may define $e^{-2\Phi_4}=e^{-2\Phi}\widehat{\mathcal V}$ before the four-dimensional Einstein-frame rescaling. A variation of volume at fixed $\Phi_4$ is not a variation at fixed ten-dimensional $\Phi$.

Type IIA

The RR potentials reduce schematically as

$$\begin{aligned} C_1&=A^0,\\ C_3&\supset A^a\wedge\omega_a +\xi^K\alpha_K-\tilde\xi_K\beta^K. \end{aligned}$$

There are no harmonic one-forms, so $C_1$ supplies no internal scalar. The $C_3$ expansion gives $h^{1,1}$ vectors and $b_3=2h^{2,1}+2$ real RR scalars. A purely external three-form has no propagating local degree of freedom in four dimensions; its constant-flux branch is set to zero here.

The total number of vectors is $h^{1,1}+1$. One combination is the graviphoton, leaving $h^{1,1}$ vector multiplets whose scalars are the $T^a$. The remaining scalars number

$$\begin{aligned} n_{\mathrm{scalar}}^{\mathrm{IIA}} &=(2h^{1,1}+2h^{2,1}+2)+(2h^{2,1}+2)\\ &=2h^{1,1}+4(h^{2,1}+1). \end{aligned}$$

Hence $n_H=h^{2,1}+1$. The extra hypermultiplet includes the four-dimensional dilaton and the scalar dual to $B_2^{(4)}$, together with RR partners. This local grouping does not assert that its kinetic metric is globally decoupled from the other hypermultiplets.

Type IIB and self-duality

The IIB RR potentials are $C_0,C_2,C_4$. The axion $C_0$ gives one scalar, while $C_2$ gives $h^{1,1}$ internal scalars and one external two-form, dual to another scalar. For $C_4$, use harmonic four-forms $\tilde\omega^a$ dual to $\omega_a$:

$$C_4\supset D_2^a\wedge\omega_a +V^K\wedge\alpha_K-U_K\wedge\beta^K +\rho_a\tilde\omega^a.$$

The candidate vectors $(V^K,U_K)$ number $2h^{2,1}+2$, but the ten-dimensional five-form is self-dual. It relates their field strengths by four-dimensional electric–magnetic duality. An electric description therefore has only $h^{2,1}+1$ independent vectors. The same self-duality relates the $D_2^a$ two-forms to the $\rho_a$ scalars; they must not be counted twice. This reduction is explicit in Grimm and Louis, section 2.

Removing the graviphoton leaves $h^{2,1}$ vector multiplets, paired with the complex-structure moduli. The RR scalar count is $1+h^{1,1}+1+h^{1,1}$, and therefore

$$\begin{aligned} n_{\mathrm{scalar}}^{\mathrm{IIB}} &=(2h^{1,1}+2h^{2,1}+2)+(2h^{1,1}+2)\\ &=2h^{2,1}+4(h^{1,1}+1). \end{aligned}$$

This gives $n_H=h^{1,1}+1$.

Both theories use the same smooth compact full-$SU(3)$ threefold. IIA vectors arise from $C_1$ and $C_3$; IIB vectors arise from $C_4$ and are halved by five-form self-duality. Each total includes one graviphoton. Scalar totals then determine the hypermultiplet counts in the flux-free $\mathcal N=2$ theory.

The two-derivative scalar space locally separates into a special Kähler vector-multiplet sector and a quaternionic Kähler hypermultiplet sector. In this ungauged setting, corrections can change kinetic metrics without generating a generic scalar potential for the neutral moduli. Fluxes, gauging, orientifold projections and added branes can change that conclusion and require a new analysis.

Mirror symmetry and the noncompact conifold

The IIA/IIB exchange on the same $X$ suggests the geometry of mirror symmetry. A mirror pair $(X,X^\vee)$ has

$$h^{1,1}(X)=h^{2,1}(X^\vee), \qquad h^{2,1}(X)=h^{1,1}(X^\vee),$$

so $\chi(X^\vee)=-\chi(X)$. The physical statement is stronger: type IIA on $X$ and type IIB on $X^\vee$ describe equivalent compactifications, with the appropriate identification of their moduli and quantum corrections. Equal Hodge data alone do not prove this equivalence. In the vector sector, mirror symmetry turns quantum Kähler information on the IIA side into complex-structure period information on the IIB side.

The next page changes from a compact internal manifold to a local noncompact Calabi–Yau singularity. The conifold is

$$\sum_{a=1}^4z_a^2=0,$$

or, after a linear change of variables, the rank-one matrix locus $\det Z=0$. Away from the origin it has complex dimension three. Its Ricci-flat metric is a cone,

$$ds_6^2=dr^2+r^2ds_{T^{1,1}}^2, \qquad T^{1,1}=\frac{SU(2)\times SU(2)}{U(1)}.$$

The metric radius satisfies $r\propto(\sum_a|z_a|^2)^{1/3}$, not the square root of that expression. Klebanov and Witten, section 2 derive this scaling and the link. A noncompact cone has different normalizability and boundary questions from a compact threefold, so the compact Hodge-counting formulas cannot simply be applied to it. D3-branes at its tip lead to the Klebanov–Witten example; the brane projection and near-horizon limit are additional ingredients.

Common pitfalls

Replacing full holonomy by a weaker condition. Ricci flatness, trivial canonical bundle and full $SU(3)$ holonomy have different logical roles. A smaller holonomy group can change both the Hodge diamond and the number of preserved supercharges.

Normalizing Ω before differentiating it. A local holomorphic section is free to rescale holomorphically. A pointwise unit normalization is generally nonholomorphic and cannot be inserted into $K_{\mathrm{cs}}$ as if nothing changed.

Counting dual fields as independent fields. An external two-form and its scalar dual are alternative descriptions. In IIB, five-form self-duality also pairs the candidate electric and magnetic vectors; the graviphoton must then be separated from the vector-multiplet count.

Treating cohomology as a complete effective action. Hodge numbers count modes under the stated hypotheses. Their kinetic couplings require harmonic-form inner products and period data, and quantum corrections or added ingredients require further analysis.

Exercises

Exercise 1. A smaller-holonomy test

The nonzero Hodge numbers of $K3$ are $h^{0,0}=h^{2,0}=h^{0,2}=h^{2,2}=1$ and $h^{1,1}=20$. An elliptic curve $E$ has all four entries $h^{0,0},h^{1,0},h^{0,1},h^{1,1}$ equal to one. Use Künneth to find $h^{1,0},h^{2,0},h^{1,1},h^{2,1},b_2,b_3$ and $\chi$ for $K3\times E$. Explain its type II supersymmetry and why the full-$SU(3)$ spectrum formula needs modification.

Solution

Künneth multiplies the Hodge polynomials, or equivalently convolves the pairs of indices. It gives

$$h^{1,0}=1, \qquad h^{2,0}=1, \qquad h^{1,1}=20+1=21.$$

For $h^{2,1}$, the contributions are $H^{2,0}(K3)\otimes H^{0,1}(E)$ and $H^{1,1}(K3)\otimes H^{1,0}(E)$, so $h^{2,1}=1+20=21$.

The second Betti number includes the nonzero holomorphic two-forms:

$$b_2=h^{2,0}+h^{1,1}+h^{0,2}=23.$$

The real cohomology calculation is equally direct:

$$b_3(K3\times E)=b_2(K3)b_1(E)=22\cdot2=44, \qquad \chi(K3\times E)=\chi(K3)\chi(E)=0.$$

The holonomy is $SU(2)$, and the $SU(4)$ spinor restricts as $\mathbf4=\mathbf2\oplus\mathbf1\oplus\mathbf1$. There are two internal singlets for each ten-dimensional parameter, twice the full-$SU(3)$ count. Type II therefore preserves sixteen real supercharges, or four-dimensional $\mathcal N=4$.

The relation $b_2=h^{1,1}$ already fails here, and the harmonic one-forms supply additional reduction modes. The numerical coincidence $b_3=2h^{2,1}+2$ and the vanishing Euler number do not restore the missing full-holonomy hypothesis.

Exercise 2. The quintic from parameters and characteristic classes

Let $X$ be a generic smooth degree-five hypersurface in $\mathbb P^4$. Use the facts that its polynomial deformations describe its complex-structure deformations and that Lefschetz gives $h^{1,1}=1$. Count the polynomial parameters modulo rescaling and projective coordinate changes. Independently compute its Euler characteristic from

$$c(TX)=\frac{(1+H)^5}{1+5H}, \qquad \int_XH^3=5.$$

Find the type IIA and IIB multiplet counts.

Solution

A homogeneous degree-five polynomial in five variables has

$$\binom{5+5-1}{5}=\binom94=126$$

coefficients. Overall multiplication does not change its zero locus, and $PGL(5)$ has complex dimension $25-1=24$. At a generic smooth point, where continuous stabilizers do not change this count,

$$h^{2,1}=126-1-24=101.$$

The supplied deformation-theoretic fact is necessary: coefficient counting by itself would not establish that every abstract deformation is polynomial.

Expanding the Chern-class expression through degree three gives

$$c(TX)=1+10H^2-40H^3.$$

Thus $c_1=0$ and

$$\chi(X)=\int_Xc_3(TX)=-40\cdot5=-200.$$

This agrees with $2(1-101)$. The Betti numbers include $b_2=1$ and $b_3=204$. Type IIA has one vector multiplet and $102$ hypermultiplets; type IIB has $101$ vector multiplets and two hypermultiplets, in each case in addition to the gravity multiplet. The calculation checks topological and spectrum data; it does not construct the Ricci-flat metric.

Exercise 3. A two-modulus metric and positivity

A smooth bicubic hypersurface in $\mathbb P^2\times\mathbb P^2$ has hypersurface class $3H_1+3H_2$. Use $\int_{\mathbb P^2\times\mathbb P^2}H_1^2H_2^2=1$ and $H_1^3=H_2^3=0$ to derive its triple intersections. For $j=xH_1+yH_2$ in the ambient positive cone $x,y>0$, calculate $G_{a\bar b}$, prove positivity, and check

$$v^aG_{a\bar b}v^b=\frac34, \qquad G^{a\bar b}K_aK_{\bar b}=3.$$

Solution

An integral over the hypersurface inserts $3H_1+3H_2$ in the ambient integral. Therefore $\kappa_{112}=\kappa_{122}=3$, while $\kappa_{111}=\kappa_{222}=0$. Including the three permutations of each mixed term gives

$$\widehat{\mathcal V}=\frac32xy(x+y), \qquad K_{\mathrm K}=-\log\bigl(12xy(x+y)\bigr).$$

The complex Hessian is one quarter of the real Hessian in $x,y$:

$$G=\frac14 \begin{pmatrix} x^{-2}+(x+y)^{-2} &(x+y)^{-2}\\ (x+y)^{-2} &y^{-2}+(x+y)^{-2} \end{pmatrix}.$$

The first diagonal entry is positive, and

$$\det G=\frac{x^2+xy+y^2}{8x^2y^2(x+y)^2}>0.$$

Sylvester’s criterion then establishes positivity throughout this cone. This is a particular geometric intersection form; it does not imply that an arbitrary cubic produces a positive metric.

Homogeneity gives a more general check. Write $K^{(v)}_a=\partial K/\partial v^a$. Since the volume is cubic,

$$v^aK^{(v)}_a=-3, \qquad 4G_{a\bar b}v^b=-K^{(v)}_a.$$

Contracting the second relation with $v^a$ gives $vGv=3/4$. Also $K_a=-iK^{(v)}_a/2$ and $K_{\bar a}=iK^{(v)}_a/2$, so

$$G^{a\bar b}K_aK_{\bar b} =\frac14(K^{(v)})^{\mathsf T}G^{-1}K^{(v)} =4v^{\mathsf T}Gv=3.$$

This last identity is often called a no-scale identity. Here it follows from the homogeneous kinetic potential; no scalar potential has been introduced in the ungauged $\mathcal N=2$ theory.

Exercise 4. A solvable complex-structure metric

Use a flat six-torus with unit real coordinate periods and

$$\Omega(\tau) =(dx^1+\tau dy^1)\wedge(dx^2+i dy^2)\wedge(dx^3+i dy^3), \qquad \operatorname{Im}\tau>0.$$

This is a geometry example with smaller holonomy, not the full-$SU(3)$ compactification used for the spectrum. Calculate $S$, the projected deformation $\chi_\tau$ and its metric. Explain the effect of $\Omega\mapsto e^{f(\tau)}\Omega$ and of a pointwise unit normalization.

Solution

Let $v=\operatorname{Im}\tau$. Direct wedge multiplication in the positive orientation gives

$$S=i\int\Omega\wedge\bar\Omega=8v, \qquad K_{\mathrm{cs}}=-\log(8v), \qquad G_{\tau\bar\tau}=\frac1{4v^2}.$$

Writing $dz^1=dx^1+\tau dy^1$, we have

$$dy^1=\frac{dz^1-d\bar z^1}{\tau-\bar\tau}.$$

Consequently,

$$\begin{aligned} \partial_\tau\Omega &=\frac{\Omega}{\tau-\bar\tau} -\frac{d\bar z^1\wedge dz^2\wedge dz^3}{\tau-\bar\tau},\\ k_\tau&=\frac1{\tau-\bar\tau}, \qquad \chi_\tau=-\frac{d\bar z^1\wedge dz^2\wedge dz^3}{\tau-\bar\tau}. \end{aligned}$$

The projected form is primitive for the product Kähler metric. Its wedge pairing divided by the $\Omega$ pairing, with the required minus sign, again gives $1/(4v^2)$.

At $\tau=i$ this also checks the Beltrami factor. The new coframe has $\mu=i\,\delta\tau/2$ after its holomorphic rescaling is separated. With $g_{1\bar1}=1/2$, metric compatibility gives $\delta g_{\bar1\bar1}=\mu$. Its contribution to the real metric-deformation norm is $|\delta\tau|^2/2=2G_{\tau\bar\tau}|\delta\tau|^2$.

A holomorphic rescaling changes $K$ by $-f-\bar f$, whose mixed derivative vanishes. In contrast, $\widehat\Omega=(8v)^{-1/2}\Omega$ has unit $S$ but is not holomorphic in $\tau$. The derivation that identified the Hessian with the projected norm assumed $\partial_{\bar\tau}\Omega=0$ in the fixed cohomology trivialization. Applying it to $\widehat\Omega$ and declaring the metric zero would violate that assumption.

Exercise 5. Electric and magnetic IIB vectors

At one point in moduli space choose a real symplectic pair of three-forms with $*_6\alpha=\beta$ and $*_6\beta=-\alpha$. Take

$$F_5=F\wedge\alpha+G\wedge\beta,$$

where $F,G$ are four-dimensional two-form field strengths. Derive the consequence of ten-dimensional self-duality and recover the IIB vector count. Why must the scalar/two-form modes of $C_4$ also be counted only once?

Solution

With product orientation $\mathrm{vol}_{10}=\mathrm{vol}_4\wedge\mathrm{vol}_6$, the product-star rule is

$$*_{10}(\eta_p\wedge\zeta_q) =(-1)^{q(4-p)}(*_4\eta_p)\wedge(*_6\zeta_q).$$

For $p=2,q=3$ the sign is positive. Therefore

$$*_{10}F_5 =(*_4F)\wedge\beta-(*_4G)\wedge\alpha.$$

Self-duality requires

$$G=*_4F, \qquad F=-*_4G.$$

These are consistent because $*_4^2=-1$ on Lorentzian two-forms. They specify one Maxwell field and its magnetic dual, rather than two independent Maxwell fields. The internal relation $*_6^2=-1$ on three-forms makes the ten-dimensional square positive.

There are $b_3/2=h^{2,1}+1$ independent electric fields. One belongs to the gravity multiplet, leaving $h^{2,1}$ vector multiplets. The local symplectic frame used in the exercise is not a claim that the Hodge star has this simple form in one fixed integral basis throughout moduli space.

The components of $F_5$ with one external and four internal indices contain derivatives of the $\rho_a$ scalars. Self-duality relates them to components with three external and two internal indices, which are field strengths of $D_2^a$. Keeping both descriptions would count each propagating degree of freedom twice. Retaining only the $h^{1,1}$ scalars gives the RR scalar total $2h^{1,1}+2$ and the hypermultiplet count derived in the page.

Exercise 6. The conifold link and its metric radius

Write the conifold as the rank-one matrix locus $Z=AB^{\mathsf T}$, with $A,B\in\mathbb C^2$. Explain the quotient by $(A,B)\mapsto(\lambda A,\lambda^{-1}B)$ and identify the fixed-radius link. A holomorphic residue three-form scales as $\Omega\mapsto\lambda^2\Omega$ under $z_a\mapsto\lambda z_a$. Use this to find the exponent of a homogeneous Kähler potential $\mathcal K\propto(\sum|z_a|^2)^\gamma$ and the relation between the algebraic and metric radii. Derive the Ricci-flat cone condition on the link.

Solution

A nonzero rank-one matrix factors into a column and a row, uniquely up to $\lambda\in\mathbb C^*$. The positive real part of this quotient can be fixed by $|A|^2=|B|^2$. At a fixed nonzero common norm, $A$ and $B$ lie on two three-spheres, and the remaining phase acts as

$$(A,B)\mapsto(e^{i\theta}A,e^{-i\theta}B).$$

The link is therefore $(S^3\times S^3)/U(1)$, of real dimension $3+3-1=5$. The rank-one locus away from its vertex has complex dimension $2+2-1=3$. This identifies the link as $T^{1,1}$; it does not identify the metric induced from the ambient flat $\mathbb C^4$ with the Ricci-flat conifold metric.

Let $\rho^2=\sum|z_a|^2$. Under positive real scaling $z_a\mapsto t z_a$, a Kähler potential $\rho^{2\gamma}$ and its Kähler form have weight $2\gamma$. Thus $J^3$ has weight $6\gamma$. The residue has weight two, so $\Omega\wedge\bar\Omega$ has weight four. Compatibility of the Calabi–Yau volume forms requires

$$6\gamma=4, \qquad \gamma=\frac23.$$

The metric radius consequently scales as $t^{2/3}$:

$$r\propto\rho^{2/3} =\left(\sum_a|z_a|^2\right)^{1/3}.$$

Finally, for $ds^2=dr^2+r^2\gamma_{ij}dy^i dy^j$ with a five-dimensional link, the additional Christoffel symbols are

$$\Gamma^r{}_{ij}=-r\gamma_{ij}, \qquad \Gamma^i{}_{rj}=\frac1r\delta^i{}_j, \qquad \Gamma^i{}_{jk}=\Gamma^i{}_{jk}[\gamma].$$

Substitution into the Ricci tensor gives

$$R_{rr}=R_{ri}=0, \qquad R_{ij}^{\mathrm{cone}}=R_{ij}[\gamma]-4\gamma_{ij}.$$

Ricci flatness requires $R_{ij}[\gamma]=4\gamma_{ij}$, the Einstein normalization used on the next page. The conifold is noncompact, so this local metric construction does not by itself define a four-dimensional compactification with the finite-volume spectrum counted above.

References

• Grimm, Thomas W., and Jan Louis. “The effective action of N=1 Calabi–Yau orientifolds.” Nuclear Physics B 699 (2004), 387–426. arXiv:hep-th/0403067. doi:10.1016/j.nuclphysb.2004.08.005. Section 2 gives the unorientifolded IIB reduction used here.
• Grimm, Thomas W., and Jan Louis. “The effective action of Type IIA Calabi–Yau orientifolds.” Nuclear Physics B 718 (2005), 153–202. arXiv:hep-th/0412277. doi:10.1016/j.nuclphysb.2005.04.007. Section 2 and Appendix A describe the unorientifolded moduli metrics and special geometry.
• Klebanov, Igor R., and Edward Witten. “Superconformal Field Theory on Threebranes at a Calabi–Yau Singularity.” Nuclear Physics B 536 (1998), 199–218. arXiv:hep-th/9807080. doi:10.1016/S0550-3213(98)00654-3.
• Yau, Shing-Tung. “Calabi’s conjecture and some new results in algebraic geometry.” Proceedings of the National Academy of Sciences 74 (1977), 1798–1799. doi:10.1073/pnas.74.5.1798.